resnet50_fold_2_v3

This model is a fine-tuned version of microsoft/resnet-50 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.2156
• Accuracy: 0.9455
• F1 Score: 0.9473
• Recall: 0.9511

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 150
• num_epochs: 100
• label_smoothing_factor: 0.15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 Score	Recall
2.7877	1.0	20	2.7748	0.2340	0.2038	0.2111
2.7713	2.0	40	2.7649	0.2821	0.1863	0.2317
2.7339	3.0	60	2.7455	0.3173	0.1730	0.2516
2.6964	4.0	80	2.7174	0.3429	0.1565	0.2663
2.6233	5.0	100	2.6713	0.3718	0.2136	0.2972
2.5505	6.0	120	2.5868	0.4103	0.2924	0.3438
2.4113	7.0	140	2.4547	0.5192	0.4524	0.4679
2.1943	8.0	160	2.2780	0.6635	0.6414	0.6308
1.9959	9.0	180	1.9887	0.7853	0.7871	0.7734
1.7676	10.0	200	1.7359	0.8365	0.8366	0.8360
1.5515	11.0	220	1.5632	0.8622	0.8634	0.8704
1.4094	12.0	240	1.4685	0.8814	0.8824	0.8885
1.3099	13.0	260	1.3924	0.8782	0.8791	0.8825
1.3021	14.0	280	1.3553	0.8942	0.8957	0.8999
1.2754	15.0	300	1.3459	0.8942	0.8965	0.9034
1.2809	16.0	320	1.3213	0.9006	0.9021	0.9064
1.2039	17.0	340	1.3051	0.9167	0.9193	0.9252
1.1827	18.0	360	1.2900	0.9135	0.9163	0.9182
1.1805	19.0	380	1.2933	0.9071	0.9092	0.9102
1.2095	20.0	400	1.2653	0.9071	0.9090	0.9072
1.1459	21.0	420	1.2834	0.9295	0.9318	0.9366
1.1535	22.0	440	1.2681	0.9231	0.9238	0.9268
1.1871	23.0	460	1.2560	0.9295	0.9310	0.9329
1.1502	24.0	480	1.2490	0.9391	0.9413	0.9440
1.1546	25.0	500	1.2599	0.9231	0.9259	0.9245
1.1360	26.0	520	1.2493	0.9295	0.9318	0.9356
1.0990	27.0	540	1.2512	0.9263	0.9284	0.9305
1.1106	28.0	560	1.2462	0.9327	0.9349	0.9385
1.1113	29.0	580	1.2399	0.9263	0.9275	0.9280
1.0854	30.0	600	1.2411	0.9231	0.9251	0.9295
1.0828	31.0	620	1.2326	0.9359	0.9383	0.9405
1.1117	32.0	640	1.2407	0.9327	0.9346	0.9351
1.0534	33.0	660	1.2415	0.9199	0.9209	0.9214
1.0891	34.0	680	1.2391	0.9231	0.9261	0.9263
1.1107	35.0	700	1.2374	0.9263	0.9264	0.9265
1.0882	36.0	720	1.2434	0.9295	0.9315	0.9302
1.0882	37.0	740	1.2353	0.9295	0.9312	0.9299
1.0987	38.0	760	1.2156	0.9455	0.9473	0.9511
1.0452	39.0	780	1.2226	0.9327	0.9348	0.9361
1.0637	40.0	800	1.2304	0.9327	0.9343	0.9344
1.0404	41.0	820	1.2234	0.9359	0.9379	0.9408
1.0338	42.0	840	1.2404	0.9263	0.9286	0.9283
1.0687	43.0	860	1.2207	0.9359	0.9372	0.9401
1.0685	44.0	880	1.2256	0.9327	0.9340	0.9371
1.0674	45.0	900	1.2268	0.9359	0.9384	0.9425
1.0527	46.0	920	1.2313	0.9327	0.9340	0.9366
1.0477	47.0	940	1.2367	0.9327	0.9339	0.9378
1.0481	48.0	960	1.2147	0.9391	0.9416	0.9457
1.0861	49.0	980	1.2289	0.9327	0.9347	0.9371
1.0387	50.0	1000	1.2225	0.9295	0.9315	0.9339
1.0592	51.0	1020	1.2252	0.9359	0.9377	0.9395
1.0397	52.0	1040	1.2287	0.9327	0.9345	0.9359
1.0772	53.0	1060	1.2241	0.9359	0.9378	0.9383
1.0461	54.0	1080	1.2280	0.9359	0.9374	0.9378
1.0345	55.0	1100	1.2358	0.9263	0.9282	0.9274
1.0456	56.0	1120	1.2268	0.9295	0.9315	0.9344
1.0747	57.0	1140	1.2241	0.9327	0.9345	0.9356
1.0194	58.0	1160	1.2096	0.9327	0.9344	0.9361
1.0270	59.0	1180	1.2059	0.9423	0.9437	0.9457
1.0273	60.0	1200	1.2210	0.9423	0.9441	0.9457
1.0271	61.0	1220	1.2130	0.9391	0.9408	0.9432
1.0133	62.0	1240	1.2107	0.9327	0.9350	0.9383
1.0046	63.0	1260	1.2163	0.9359	0.9381	0.9420
1.0196	64.0	1280	1.2262	0.9359	0.9381	0.9420
1.0338	65.0	1300	1.2231	0.9359	0.9377	0.9395
1.0166	66.0	1320	1.2208	0.9327	0.9347	0.9371
1.0027	67.0	1340	1.2165	0.9295	0.9314	0.9346
1.0150	68.0	1360	1.2331	0.9391	0.9416	0.9432
1.0003	69.0	1380	1.2159	0.9327	0.9358	0.9383

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 5.0.0
• Tokenizers 0.22.2